UNIVERSITA’ DEGLI STUDI DI VERONA
Department of Neurosciences, Biomedicine and Movement Sciences Graduate School of Health and Life Sciences
DOCTORAL PROGRAM IN
NEUROSCIENCE, PSYCHOLOGICAL AND PSYCHIATRIC SCIENCES CYCLE XXXI – 2015-18
PhD Dissertation
THE PLACEBO EFFECT IN THE MOTOR DOMAIN:
A NEURAL AND BEHAVIORAL APPROACH
Coordinator: Prof. Leonardo Chelazzi Supervisor: Prof. Mirta Fiorio
2
La verdadera ciencia enseña, por encima de todo, a dudar y a ser ignorante. (True science teaches, above all, to doubt and be ignorant.)
3 CONTENTS
PREFACE ... 5
PART I: BACKGROUND ... 10
Definition of the placebo effect ... 10
Cognitive mechanisms of the placebo effect ... 11
Neural correlates of placebo analgesia ... 12
The placebo effect in the motor domain ... 15
General aims ... 15
PART II: NEURAL INVESTIGATION OF THE PLACEBO EFFECT IN THE MOTOR DOMAIN ... 16
Current knowledge on the neural correlates of the motor placebo effect ... 16
On the role of the dorsolateral prefrontal cortex ... 20
Transcranial direct current stimulation (tDCS) ... 22
Methods ... 25
Results ... 34
Discussion ... 42
PART III: ENLARGING THE BEHAVIORAL INVESTIGATION OF THE PLACEBO EFFECT IN THE MOTOR DOMAIN ... 50
Current knowledge on the behavioral aspects of the motor placebo effect ... 50
The placebo effect on balance control ... 53
Methods ... 55
Results ... 60
Discussion ... 65
The placebo effect on motor learning ... 70
Method ... 72
4 Discussion ... 88 GENERAL CONCLUSION ... 94 ACKNOWLEDGEMENT ... 97 LIST OF PUBLICATIONS ... 98 REFERENCES ... 99
5 PREFACE
The placebo effect is a beneficial outcome that follows the administration of a treatment and that is not to be ascribed to active ingredients but to the words, contexts and beliefs that surround the treatment and that can induce psychological and neuronal changes in the recipient’s brain (Benedetti et al., 2011). This psychobiological phenomenon represents a good model to study the mind-brain-body interaction.
During the past decades, the placebo effect has attracted the interest of researchers from different backgrounds, such as psychology, neurobiology, cognitive neuroscience among others. Many important studies in the healing context, adopting clinical trials or experimental models of pain, have allowed to achieve a deep understanding of the neurobiological correlates and cognitive mechanisms involved in this phenomenon. In more recent years, however, it has become always clearer that the placebo effect is a pervasive phenomenon that extends beyond the healing context (Pollo et al., 2011). With regard to this, different lines of evidence have shown that the placebo effect can be found in many contexts, such as the cognitive, the sensory, the emotional and the motor domains (Beedie & Foad, 2009; Beissner et al., 2015; Schienle et al., 2013; Schwarz & Büchel, 2015).
During my Ph.D. I have been particularly interested in enlarging our knowledge on the placebo effect in the motor domain. This interest derives from scientific curiosity, as well as from the potential future translational impact of the motor placebo effect in sports and pathology. For instance, it could be possible to think at the placebo effect as a strategy to implement the outcome obtained with the traditional sport trainings and also as a complementary strategy for motor recovery, for instance in patients in whom the pharmacological treatment is less effective. However, before achieving this translational impact, some issues need to be clarified: First, the neural correlates of the placebo effect in the motor domain are still largely unknown; Second, knowledge is still limited on the type of motor functions that could be influenced by the placebo effect.
With regard to the first issue, up-to-now very few studies have investigated the neural correlates of the motor placebo effect in healthy participants. These studies highlighted the role of the primary motor cortex (M1) and the supplementary motor
6
area (SMA) (Fiorio et al., 2014; Piedimonte et al., 2015). Different approaches in patients affected by movement disorders, like Parkinson’s disease, hinted at the involvement of subcortical structures, like the subtalamic nucleus (STN), the substantia nigra pars reticulata (SNr), the ventral anterior (VA) and the anterior ventrolateral (VLa) nuclei of the thalamus (Benedetti et al., 2004; Benedetti et al., 2009). All these brain areas are not isolated, but are strongly connected with other brain regions, such as the dorsolateral prefrontal cortex (dlPFC) (Hasan et al., 2013; Mayberg et al., 2002; Miller & Cohen, 2001), the cingulate cortex (Asemi et al., 2015; Mayberg et al., 2002; Petrovic et al., 2002), or the orbitofrontal cortex (Petrovic et al., 2002) among others. Of note, these areas have been consistently shown to be involved in placebo analgesia (Ashar et al., 2017; Wager & Atlas, 2015) and it is reasonable to hypothesize that they could also play a role in the motor placebo effect. In particular, the dlPFC plays a crucial role in placebo analgesia and it is also involved in higher-order cognitive functions, like expectation and anticipation, that are at the basis of the placebo effect. Moreover, the dlPFC also has connections with motor brain areas, thus suggesting a potential role of this brain region in the motor placebo effect. Part of my Ph.D. project was dedicated to tackle the role of the dlPFC in the motor placebo effect in healthy participants. In Part II of this thesis, I will describe a series of three experiments in which we applied non-invasive brain stimulation combined with a placebo procedure on force production. This investigation has allowed to enlarge the current knowledge on the neural correlates of the placebo effect in the motor domain, by demonstrating that the dlPFC could also play a role, especially when expectation is the main cognitive mechanism at the basis of the placebo effect.
With regard to the second issue, so far, the behavioral investigation of the placebo effect has addressed some dimensions of motor performance, such as force production, movement speed and resistance to fatigue (Beedie & Foad, 2009; Fiorio et al., 2018; Pollo et al., 2011). The potential effects of placebo on other crucial motor functions remains unknown. Motor performance is a complex definition that embraces different dimensions, such as precision control, balance, visuomotor coordination, motor sequence learning and motor adaptation, among others. Some of these motor functions are important not only in sports, but also in daily life
7
activities. Hence, extending the behavioral investigation of the placebo effect on other motor functions may help to achieve a better understanding of the placebo effect itself, as well as to enlarge its range of application to daily life activities. During my Ph.D., I have tried to explore whether a placebo procedure can improve two relevant motor functions present in our daily life, like balance control and motor sequence learning. Balance can be defined as the capacity “to control the body’s position in space for stability and orientation” (O’Sullivan, 2007). Balance control is crucial to accurately perform most of the activities in daily life, such as getting up from bed, walking, waiting on a queue or simply having a shower. Conversely, disturbances in balance control, such as those present in Parkinson’s disease or in the elderly population can lead to higher risk of fall, which limits the quality of life (Jacobs et al., 2005; Maki et al., 1994; Pfortmueller et al., 2014). In Part III of this thesis, I will describe a new protocol that we developed to improve balance control in healthy participants with a placebo procedure. We think that extending the potential beneficial effects of placebos on balance control may allow to provide in the future new strategies for the rehabilitation of gait disorders for which the pharmacological treatment is often not effective.
Motor skill learning in another important motor function that permits to convert isolated and specific movements into well-performed skills through practice (Dayan & Cohen, 2011; Wolpert et al., 2011). In particular, motor sequence learning is crucial in many tasks, such as cooking and cleaning, and to acquire skills, like writing or cycling that are present during the lifespan. In Part III of this thesis, I will describe a study that we conducted to investigate the placebo effect on the learning of motor sequences in healthy participants. Even in this type of study, we envisage a potential future impact for rehabilitative trainings after an injury (i.e. stroke) (Kitago & Krakauer, 2013).
To summarize, with this work we have tried to expand the current knowledge on the placebo effect in the motor domain from two different perspectives: from one hand, we have conducted a neural investigation on the involvement of a higher-order brain region (the dlPFC) in the motor placebo effect; from the other hand, we have enlarged the behavioral investigation of the placebo effect to two different movement functions: balance control and motor sequence learning.
8 ABSTRACT
The placebo effect is a fascinating psychobiological phenomenon that allows to investigate the mind-body interaction. It is typically induced by the application of an inert treatment along with verbal suggestion of beneficial outcomes. The placebo effect has been deeply investigated in the field of pain, although different lines of evidence suggest that it is also present in other domains, like the motor domain. Extending our knowledge of the placebo effect in the motor domain can have important future translational impacts in sports and pathology. The aim of my PhD project was to study the placebo effect in the motor domain at two different levels: the neural and the behavioral level.
Regarding the neural level, knowledge on the brain regions related to placebo effect in the motor domain is limited. We aimed at filling in this knowledge gap by investigating the role of the dlPFC, a brain region also involved in placebo analgesia. The dlPFC elaborates expectation, a cognitive function at the basis of the placebo effect and shares some connections with other brain regions involved in motor control. Hence, there are many clues to hypothesize a role of the dlPFC in the motor placebo effect. To tackle this issue, three different experiments were conducted in which the dlPFC was stimulated by means of transcranial direct current stimulation (tDCS) together with a placebo procedure on force production. We found that the left dlPFC is involved in the expectation-induced enhancement of force, specifically in those subjects who respond to the placebo effect (placebo-responders).
Regarding the behavioral level, it should be noticed that many behavioral studies have shown that the placebo effect can enhance different aspects of motor performance associated to sports, such as force, speed or endurance. It is still unknown, however, whether the placebo effect can also improve other motor functions, important for many daily life activities, like balance or motor sequence learning. Thus, another objective of my PhD was to investigate the potential influence of the placebo effect on two motor functions that are closer to daily life activities. To this aim, a first study was conducted to understand whether balance control, a motor function needed for many daily life activities and for preventing falls, could be enhanced in healthy participants by a placebo procedure consisting
9
of verbal suggestion. We found that different parameters of balance (in the three-dimensional space and in the medial-lateral direction) and the subjective perception of stability were improved by the placebo procedure.
A second behavioural study was run to investigate whether the application of a placebo treatment consisting of verbal suggestion could help in improving motor sequence learning. In this case, we also aimed to tackle a differential role of two types of placebo treatments: one motor and one cognitive. The motor placebo procedure consisted of transcutaneous electrical nerve stimulation (TENS) applied the hand muscles involved in the task together with verbal information on the beneficial effects on muscle activity. The cognitive placebo procedure consisted of sham transcranial direct current stimulation (tDCS) applied over the frontal region together with verbal information on the beneficial effects on attention. Our findings did not show a clear improvement of performance following the placebo procedures, but a significant effect on the subjective perception of fatigue. More precisely, while the placebo procedure directed to the motor function (TENS) could reduce the perception of physical fatigue, the placebo procedure focused on cognitive functions (sham tDCS) could decrease the perception of both mental and physical fatigue.
Altogether these investigations represent an attempt to deepen our understanding of the neural correlates of the motor placebo effect and to enlarge the potential behavioural influence of placebos on different motor functions.
10 PART I: BACKGROUND
Definition of the placebo effect
Placebo is a latin word that means “I shall please” and it was first documented in the Vespers for the Dead (Psalm 116, 9th verse) as Placebo Domino in regione vivorum (Hart, 1999; Jacobs, 2000). By that time, the term placebo started to be used with the notion of “pleasing”. For example, some mourners were hired to “sing placebos” to adulate the dead at the burials (Kerr et al., 2008).
Some years had to elapse to find the word placebo associated to the medical lexicon. During the late 18th century, a British physician used the word placebo for describing a method to give comfort and please to patients with illnesses without cure (Kerr et al., 2008), thus associating the word to console or relieve more than to cure. Several years later, in 1811, a new explanation of the word placebo was registered and defined in the Hooper’s Medical Dictionary. Placebo was defined as “any medicine adapted more to please than benefit the patient” (Kerr et al., 2008; Finniss, 2018).
During the 20th century, the term placebo started to be slightly transformed and associated to therapeutic rituals and deception (Carlill, 1918). But it was during the mid-20th century when the word placebo started to be associated with the placebo effect. In 1955, Henry K. Beecher discovered the effect of placebo by using a placebo-controlled double-blinded design, using the placebo as a tool to dissociate the real pharmacological effect from the suggestion that arises when a pharmacological treatment is applied. In his article, the placebo resulted effective in the 35.2% of the cases and the effect was associated to the subjective component that emerges from the applied treatment. Several subsequent researches have attempted to obtain a better definition of the placebo effect (Benedetti, 2002; Shapiro, 1997; Vase, 2002).
A related aspect to consider is the difference between the terms placebo effect and placebo response that are usually considered interchangeable in the literature, although a difference exists between these two terms. According to Hoffman et al. (2005), the term placebo response refers to the individual change that is caused by
11
a placebo manipulation or treatment simulation, while the term placebo effect refers to the average enhancement that occurs in a group of subjects after receiving a placebo manipulation or treatment simulation.
Today, the most accepted definition of the placebo effect refers to a psychobiological phenomenon that could be defined as a physical or psychological benefit following the administration of an inert substance (or sham treatment) together with a positive context inducing positive expectations about its effect (Benedetti et al., 2016).
Cognitive mechanisms of the placebo effect
Placebo responses are the result of a complex interaction of different biopsychosocial factors, such as cognitive functions like expectation and learning, personality traits and genetic factors (Colagiuri et al., 2015; Colloca et al., 2014; Colloca & Miller, 2011; Corsi & Colloca, 2017; Peciña et al., 2013). In particular, expectation and learning represent the main cognitive mechanisms implicated in the modulation and formation of the placebo effect.
The expectancy model is considered as the central mechanism for the development of the placebo effect and makes a special mention of the importance role of verbal suggestions (Colloca & Miller, 2011; Kirsch, 2018). According to the Theory of Expectancy (Kirsch, 1985), the previous belief of a person about what will happen in certain circumstances will determine what that person will experience in the end. Thus, a response (placebo) can occur as consequence of expecting a positive outcome. In other words, the expectation can be generated by positive verbal suggestions associated with a treatment (actually inert), thus inducing a positive expectation and creating a real effect. This induction of positive expectation about the effect of an ergogenic aid (actually inert) in reducing symptoms (i.e., pain) can be sufficient to modify pain perception, pain sensation and dopamine release (Benedetti et al., 2003) and can also modulate motor and cognitive performance (Beedie et al., 2009; Colagiuri et al., 2011).
Another crucial psychological mechanism to induce the placebo effect is learning. It has been demonstrated that different types of conditioning, such as classical conditioning or instrumental conditioning, are responsible for the formation of the
12
placebo effect (Benedetti et al., 2003; Colloca & Miller, 2011; de la Fuente-Fernández et al., 2009). Classical conditioning in the placebo field, has been proposed as the primary method to produce learning (Colloca et al., 2010; Montgomery & Kirsch, 1997; Voudouris et al., 1990). In this case, a previous experience of benefit associated with the exposure of a real treatment effect can later turn out to be beneficial when the real treatment is replaced by an inert (similar) treatment. The length of the learning process is a determinant factor for the placebo effect, in that a longer learning period during the placebo manipulation or treatment simulation resulted in stronger placebo response (Colloca et al., 2010). Interestingly, another type of learning, such as social observation, can produce placebo effects (Colloca & Benedetti, 2009; Schenk et al., 2017). Specifically, a person could learn how to respond to a condition by just observing the beneficial effect of a demonstrator after the application of a treatment.
These two-main cognitive mechanisms, expectation and learning, can interact to generate and maintain placebo effects (Ashar et al., 2017). In agreement with a cognitive interpretation of conditioning (Reiss, 1980; Rescorla, 1988), reinforced expectations can be induced after a conditioning procedure. In this way, after repeated exposure to the effects of a treatment, the individual knows what to expect when the treatment is applied again (Kirsch, 2018). Evidences converge in indicating that the placebo effect induced by the combination of both verbally-induced expectancy and conditioning results in stronger effects (Colloca et al. 2008; Schafer et al. 2015) than the placebo effect induced by the application of only verbal suggestion (Colloca et al., 2008; de Jong et al., 1996) or only conditioning (Montgomery & Kirsch 1997). Up to now, these evidences support the idea that recurrent positive experiences along with a cognitive ascription of benefit to the treatment is the best combination to obtain strong placebo effect (Ashar et al., 2017).
Neural correlates of placebo analgesia
Many studies have investigated the neural underpinnings of the placebo effect in pain. As we know, different cognitive mechanisms can induce a placebo response. Thus, the placebo effect cannot be supported by a simple brain mechanism or
13
system (Ashar et al., 2017). In Benedetti’s words, “there is not a single placebo effect but many” (Benedetti, 2006). Therefore, different neural mechanisms could be also engaged depending on the function on which the placebo effect works (for instance, pain). Understanding the numerous neurobiological mechanisms that are involved in different placebo responses can help to understand the complex mind-brain-body interaction (Benedetti, 2006).
A considerable number of researches have explored the neural correlates of the placebo effect with different modern techniques such as electroencephalography (EEG), magnetoencephalography (MEG), functional magnetic resonance imaging (fMRI) or positron emission tomography (PET). It is important to note that most of the studies have focused on placebo analgesia (Ashar et al., 2017; Wager & Atlas, 2015). There is clear evidence that placebo procedures can significantly reduce pain-related responses occurring in the pain-processing system (Wager & Atlas, 2015). A clear and direct correlation was demonstrated between high placebo responses and reduced activity in pain-processing systems (Wager et al. 2007) or brain areas such as the dorsal anterior cingulate cortex, the thalamus and the mid- and anterior insula among others (Geuter et al., 2013; Wager & Atlas, 2015; Watson et al., 2009). For example, the responses related to pain that occur in the somatosensory areas and in the behavior are due to the network that connects the anterior cingulate cortex and the periaqueductal grey (Lui et al., 2010; Wager et al., 2004). Interestingly, the placebo effect is not related only with the central components of pain. Brain regions involved in higher-order cognitive functions, like anticipation of benefit and expectation play also a role in the placebo effect. Some of these brain areas show activation before and during painful stimulation, like the dlPFC, the ventromedial prefrontal cortex or the mid-lateral orbitofrontal cortex (Kong et al., 2006; Wager & Atlas, 2015; Watson et al., 2009). Moreover, the increased activity of the reported brain areas directly correlates with the amount of reported pain reduction (Wager & Atlas, 2015) (Figure 1).
14
Figure 1. Neural correlates of placebo analgesia. A general view of some brain areas
involved in the placebo analgesic effect. The areas that are represented in orange are involved in pain perception and show reduced activation after a placebo treatment. Some of these areas are the thalamus (Thal), periaqueductal grey (PAG), dorsal anterior cingulate cortex (dACC) and secondary somatosensory (S2) among others. Moreover, the areas that are represented in green are also related to other higher-order functions, like expectation or maintenance of context information. These areas show increased activation before or during a placebo treatment. These areas include the dorsolateral prefrontal cortex (dlPFC), the ventromedial prefrontal cortex (vmPFC), the lateral orbitofrontal cortex (lOFC) and periaqueductal grey (PAG) among others. Adapted from Wager & Atlas, 2015.
The placebo effect can also modify the release of different neurotransmitters. It has been discovered that the administration of naloxone, an opioid antagonist, could block the effect of placebo, thus demonstrating the involvement of the endogenous opioid system in placebo analgesia (Levine et al., 1978). From this first experiment, several studies have characterized the placebo effect in pain using naloxone, like reduction of heart rate or decrease β-adrenergic response (Colloca & Benedetti, 2005: Pollo et al., 2003).
15 The placebo effect in the motor domain
The investigation of the placebo effect in the motor domain is not new, going back to the 70’s when Ariel and Saville (1972) performed a study on weightlifters and noticed that athletes who thought to have taken an ergogenic aid (actually a placebo) improved the performance. Since then, many other behavioral studies demonstrated the powerful effect of placebo in many sports (Beedie & Foad, 2009; Pollo et al., 2011). More recently, the investigation was enlarged to experimental models of motor performance in laboratory settings both in athletes and non-athletes (Benedetti et al., 2007; Carlino et al., 2014; Kalasountas et al., 2007; McKay et al., 2012). Finally, the interest on this field has become always wider to include also the neurophysiological investigation on the neural bases of the effect.
I will describe in more detail some of these studies in the following parts of my thesis, since my main interest was to study different levels of the placebo effect in the motor domain.
General aims
All the studies on placebo analgesia were very important in starting the experimental investigation on the neural and cognitive mechanisms of the placebo effect. As anticipated in the Preface, my interest was to export this investigation to the motor domain. In this regard, some questions still remain unanswered and during my Ph.D., I tried to design some studies to answer at least in part some of these questions. In particular, one goal of my research was to understand the role of a frontal brain area (like the dlPFC) in the placebo effect in the motor domain (Part II). Furthermore, I tried to enlarge the investigation of the behavioral aspects of the motor placebo effect by tackling two motor functions that are crucial in daily life activities (and not only in sports, as mainly investigated so far), like balance control and motor sequence learning (Part III).
16
PART II: NEURAL INVESTIGATION OF THE PLACEBO EFFECT IN THE MOTOR DOMAIN
Current knowledge on the neural correlates of the motor placebo effect The neural correlates of the placebo effect in the motor domain are still largely unknown. So far, only two studies in healthy subjects have shown the contribution of some cortical brain areas in the placebo effect in the motor domain (Fiorio et al., 2014; Piedimonte et al., 2015). Fiorio et al. (2014) investigated whether the primary motor cortex (M1) could contribute to the behavioral increase of force production after a placebo procedure. To this end, the Authors used single-pulse transcranial magnetic stimulation (TMS) over M1 to evaluate the excitability of the corticospinal system. Motor evoked potentials (MEP) and cortical silent period (CSP) could be recorded with surface electrodes positioned over the muscle involved in the force task (i.e., the first dorsal interosseous). The amplitude of the MEP served as indirect measure of cortical and spinal motor circuits activity (Rösler & Magistris, 2008) and the duration of the CSP represented activity of inhibitory circuits at the cortical level (Wolters et al., 2008). Healthy subjects were randomized in four different groups: only verbal suggestion (I); verbal suggestion and conditioning (II); control with inert treatment (III) and control without inert treatment (IV). Participants were required to perform a motor task consisting of an abduction movement of the right index finger against a piston. They were instructed to press as strongly as possible and could see in real-time a visual feedback of the exerted force. Subjects of group I and II (experimental groups) underwent a placebo procedure consisting in the application of an inert electrical device (transcutaneous electrical nerve stimulation, TENS) over the muscle involved in the task together with positive verbal suggestion of improvement. Additionally, participants of group II received a conditioning procedure consisting of a surreptitious increase of the visual feedback of force. Groups III and IV served as control; with the former receiving the same TENS application but with overt verbal information that it was inert and the latter performing the same motor task but without TENS application. To explore the involvement of M1 in the placebo effect, single-pulse TMS was
17
delivered when all the participants exerted the same amount of force (i.e., 30% of the maximum voluntary force). Results showed higher MEP amplitudes after the placebo procedure in both experimental groups (I and II). Furthermore, the duration of the CSP was reduced only in the group who received verbal suggestion and conditioning (II). These findings suggest that the excitability of the corticospinal system can be modulated by a placebo procedure and results in higher force production.
The second evidence on the neural correlates of the motor placebo effect in healthy participants derives from an EEG study by Piedimonte et al. (2015). The study investigated the effect of a placebo procedure in reducing fatigue in a strength task. A specific component, called readiness potential (RP), was extracted from the EEG signal. The RP is associated to the preparation of voluntary movements (Shibasaki & Hallett, 2006) and is modulated by fatigue (Schillings et al., 2006; Slobounov et al., 2004). Moreover, the supplementary motor area (SMA) and M1 have been proposed as brain sources of the RP (Deecke, 1996; Shibasaki & Hallett, 2006). The task consisted of a repetition of flexion movements for lifting a weight with the index finger to induce fatigue. Participants were divided in two groups, placebo and control. The placebo group had to ingest a substance (actually inert) along with verbal suggestion that it was caffeine and could reduce fatigue. The control group, instead, did not received any treatment. Results showed that while in the control group there was higher perception of fatigue and higher RP amplitude after the procedure, in the placebo group perception of fatigue was reduced and the RP amplitude did not increase through the experiment. According to the Authors, a central mechanism could play a role in the placebo-induced decrease of fatigue, before movement execution (Piedimonte et al., 2015).
A different approach to investigate the neural correlates of the placebo effect in the motor domain derives from patients with motor symptoms, like Parkinson’s disease. Parkinson’s disease is a neurodegenerative disorder of the basal ganglia. A reduction of dopaminergic neurons in the substantia nigra is the cause of the occurrence of motor symptoms (Opara et al., 2017). A direct and straightforward comparison of the neural correlates of the motor placebo effect of parkinsonian patients with healthy individuals may be risky and a matter of bias. Nonetheless,
18
the results obtained from studies in parkinsonian patients can give important information on the neural correlates of the motor placebo effect. Parkinson’s disease allows to explore how subcortical structures, like the basal ganglia, are related to cognitive and motor functions (Mallet et al., 2007).
Different studies have demonstrated that the dopaminergic system can be modulated by a placebo procedure. Specifically, de la Fuentes-Fernández et al. (2001 and 2002) made a positron emission tomography study to evaluate the amount of raclopride, an antagonist of dopamine receptors, after a placebo procedure in parkinsonian patients. The Authors found a reduced amount of raclopride in both the dorsal and the ventral striatum after placebo administration, suggesting that the placebo effect could be associated to the release of endogenous dopamine in subcortical structures (de la Fuentes-Fernández et al., 2001; de la Fuentes-Fernández et al., 2002). In another study (Lidstone et al., 2010), the amount of dopamine release in the striatum was modified depending on the told probability to receive an active treatment (actually a placebo). Precisely, patients who though to had 75% or 100% probabilities of receiving the treatment showed a significant increase of dopamine release, while those who though to had 25% or 50% of receiving the treatment did not show any change. The study demonstrated that the dopamine release in the striatum is related to the expectation of benefit subsequent to a placebo treatment (Lidstone et al., 2010).
The surgery to implant deep brain stimulation (DBS) in parkinsonian patients allows to evaluate the activity of neurons of the stimulated area. Electrophysiological studies showed the involvement of the subthalamic nucleus (STN) in the placebo effect. Benedetti et al. (2004) investigated the firing rate of the neurons of the STN after a placebo procedure in selected patients who were waiting for neurosurgical intervention with DBS. Once selected, patients underwent a conditioning procedure consisting of the administration of apomorphine (dopamine agonist) before the surgical intervention. During the surgery, injection of saline solution along with verbal suggestion of motor improvement was applied to the patients. The placebo procedure was applied during the electrophysiological recording of the neural activity in the STN. Benedetti et al. (2004) demonstrated that after a placebo procedure, some patients (responders) showed a reduction of
19
muscular rigidity and also a reduced discharge frequency and evoked non-bursting activity in the STN.
To deeper understand the potential changes of other subcortical structures in the motor placebo effect, Benedetti et al. (2009) recorded the activity of the substantia nigra pars reticulata (SNr), the ventral anterior (VA) and the anterior ventrolateral (VLa) nuclei of the thalamus. All these structures are involved in motor control and connected to the STN. Using a similar paradigm, they found a significantly lower activity in the STN and the SNr, whereas neuronal activity was higher in the VA and VLa nuclei only in patients who showed a reduced muscle rigidity. Moreover, a recent study demonstrated that to obtain clinical and neural changes (in the thalamus) a conditioning procedure should be used in which a real drug is administered together with the placebo procedure (Benedetti et al., 2016).
As we have been observing above, different cortical and subcortical structures (M1, SMA, STN, SNr, VA and VLa) are involved in the placebo effect in the motor domain and, interestingly, these structures are connected with other brain areas, such as the dlPFC (Hasan et al., 2013; Mayberg et al., 2002; Miller & Cohen, 2001), the cingulate cortex (Asemi et al., 2015; Mayberg et al., 2002; Petrovic et al., 2002), or the orbitofrontal cortex (Petrovic et al., 2002) which in turn have also been related to placebo analgesia (Ashar et al., 2017; Wager & Atlas, 2015). Hence, these evidences emphasize the idea that some of these areas could also have a role in the motor placebo effect. Among all these brain regions, the dlPFC seem to be a suitable candidate, due to it has a relevant role in placebo analgesia and in higher-order cognitive functions, like expectation and anticipation, which are considered the basis of the placebo effect. Remarkably, it has been demonstrated that it communicates with motor brain areas, thus indicating a potential role in the motor placebo effect.
In particular, this second part of my thesis describes a series of three experiments conducted to investigate the involvement of the dlPFC in the motor placebo effect. To this purpose, we proposed a motor task to measure the subjects’ force level before and after a placebo procedure and modulated the activity of the dlPFC by means of non-invasive brain stimulation with transcranial direct current stimulation (tDCS). The selection of a force task among all the other motor task (i.e. movement
20
speed or resistance to fatigue) was due to one main reason. That is, a previous study has demonstrated the paradigm on the neural correlates of force during a placebo procedure (Fiorio et al., 2014). Thus, supporting by this study, we could investigate whether other brain areas involve in force (in our case the dlPFC) could also be modulated by a placebo procedure.
On the role of the dorsolateral prefrontal cortex
So far, a very limited number of studies attempted to investigate the potential neural correlates of the placebo effect in the motor domain. As mentioned previously, only two studies tackled this issue in healthy participants. More precisely, a first study has shown that the placebo-induced enhancement of force was related to an increase of activity in the left M1 (Fiorio et al., 2014). This was possible thanks to the application of TMS in healthy participants, which allowed to observe an enhanced amplitude of the motor evoked potentials and a shortening of the cortical silent period after the placebo procedure (Fiorio et al., 2014). A following EEG study showed that the placebo-induced decrease of fatigue was associated to a stable amplitude of the readiness potential (Piedimonte et al., 2015). The readiness potential is interpreted as the anticipatory phase of a movement and arises from the supplementary motor area and M1 (Deecke, 1996; Shibasaki & Hallett, 2006). Hence, the two studies in healthy participants indicate that two cortical areas (M1 and SMA), involved in movement execution and preparation play a role in the motor placebo effect.
Nonetheless, a more complex brain network is necessary for the cognitive control of motor behavior. The dlPFC together with other frontal regions, plays a prominent role in this complex network (Hasan et al., 2013; Miller & Cohen, 2001). It is worth mentioning that several studies have demonstrated that the dlPFC has an important role in placebo analgesia (Egorova et al., 2015; Geuter et al., 2013; Kong et al., 2006; Krummenacher et al., 2010; Lui et al., 2010; Peciña et al., 2013; Wager et al., 2004; Watson et al., 2009) and in elaborating expectation. As we know from the literature, expectation is one of the crucial cognitive mechanisms at the basis of the placebo effect (Krummenacher et al., 2010; Jubb & Bensing, 2013). Consequently, it is rational to conjecture that the dlPFC could be a potential actor in the placebo
21
modulation of motor performance. Our goal was to explore this hypothesis. To this end, we decided to apply the tDCS over the dlPFC during a placebo procedure in the motor domain, and precisely on a force task. The control of force requires the activation of both cortical and subcortical brain regions, like the prefrontal cortices, the cingulate motor area, premotor area, pre-SMA, SMA, the cerebellum and the basal ganglia, as we know from neuroimaging studies (Badoud et al., 2017; Dettmers et al., 1995; Ehrsson et al., 2000; Neely et al., 2013; Schmitz et al., 2005; Vaillancourt et al., 2007; Wasson et al., 2010). Several of these brain areas are connected with the dlPFC and they could exert a potential top-down control on force through the activity of the dlPFC (Alexander et al., 1986; Bates & Goldman-Rakic, 1993; Cieslik et al., 2013; Lu et al.,1994; Miller & Cohen, 2001; Schmahmann & Pandya, 1997; Petrides & Pandya, 1999). Additionally, the dlPFC is also involved in the selection of the quantity of force to be applied (Vaillancourt et al., 2007) and in the prediction of force amplitude (Wasson et al., 2010), supporting a direct role of the dlPFC in the cognitive control of force. As far as we know, no study so far has tackled the role of the dlPFC in the placebo effect on force.
Although there is clear evidence on the role of the dlPFC on placebo analgesia, there is not a definite role on the hemisphere (whether the left or the right dlPFC). Some studies reported that the left dlPFC is involved in placebo analgesia (Peciña et al., 2013; Watson et al., 2009), whilst other studies suggested the involvement of the right dlPFC (Egorova et al., 2015; Lui et al., 2010). Finally, other studies have demonstrated that the left and right dlPFC could act bilaterally in placebo analgesia (Kong et al., 2006; Krummenacher et al., 2010; Wager et al., 2004). It could be speculated that the type of experimental protocol and procedure adopted could partly explain the different results (Egorova et al., 2015; Lui et al., 2010). In our study, subjects had to perform the motor task with the right hand, requiring the involvement of the contralateral (left) primary motor cortex. Hence, in our study we hypothesized that the left dlPFC could be involved in the placebo-induced enhancement of force. The role of the left dlPFC was evaluated in three separate experiments and we implemented a within-subjects design in which anodal, cathodal and sham tDCS was applied to the same participant in three different days.
22
Transcranial direct current stimulation (tDCS)
We decided to apply non-invasive brain stimulation, in particular the tDCS, during a placebo procedure to investigate the left dlPFC. It may be worth describing the basic principles of tDCS before continuing with the study description. This little introduction about the tDCS helps to better clarify the reasoning behind the application of this technique together with a placebo procedure in the motor domain.
Non-invasive brain stimulation (NIBS) techniques, like repetitive transcranial magnetic stimulation (rTMS) and transcranial electrical stimulation (tES) are widely used as a research tool for studying the human motor and cognitive functions (Brunoni et al., 2012; Jacobson et al., 2012; Pascual-Leone et al., 1994; Perceval et al., 2016; Wassermann et al., 1998). Specifically, tES allows to modulate the state of the cerebral cortex by applying a very low electrical current over the scalp. Among all the existing types of tES, transcranial direct current stimulation (tDCS) has become the most widely known, investigated and used technique to investigate cognitive processes (Antal et al., 2014; Bestmann et al., 2015; Santarnecchi et al., 2015).
Back to the 60s, Bindman et al. (1964) demonstrated that the stimulation of the rat’s cortex with subthreshold direct current induced a change of the cortical activity in a polarity-dependent manner (Bindman et al.,1964). Thirty years later, this effect was proven also in the human brain (Nitsche & Paulus, 2000; Priori et al., 1998). TDCS consists of a low-intensity direct current (1 to 2 mA) applied over the scalp by means of a pair of rubber electrodes (typically 25 or 35 cm2) inserted in sponges that are soaked in saline solution (Stagg & Nitsche, 2011). TDCS can induce neural changes in the cortical activity, functional connectivity and metabolite concentrations and in this way, it can modulate human behavior (for a review, Fertonani & Miniussi, 2017; Nitsche et al., 2015). Thus, tDCS can temporarily interfere with the activity of a cortical network and in the meantime, it is possible to test the effects of this interface on specific tasks. Different motor and cognitive functions have been shown to be modulated by the application of tDCS on specific brain regions (for a review, Perceval et al., 2016; Shin et al., 2015).
23
Regarding the type of stimulation, tDCS can be applied in two active ways, anodal and cathodal stimulation. Concerning anodal stimulation, the anode (positive electrode) is placed over the target brain area under investigation whereas the cathode (negative electrode) is positioned over a reference area. With regards to cathodal stimulation, the electrode position is reversed, so that the cathode is placed over the target brain area of interest and the anode over the reference one. Furthermore, tDCS allows the application of an inactive condition, the so-called sham stimulation, which can emulate similar sensations on the skin like the active condition, but without any cortical modulation (Gandiga et al., 2006; Nitsche et al., 2008). Most of the studies indicate that anodal stimulation increases cortical excitability while cathodal reduces it. However, with regards to the cognitive domain, the inference anodal equals improved performance and cathodal equals impaired performance is not always possible. Namely, factors like the intensity of stimulation, the duration of stimulation or the cortical activation state at the time of the stimulation can produce different unexpected outcomes (Batsikadze et al., 2013; Monte-Silva et al., 2013; Silvanto et al., 2008). Thus, it is important to consider tDCS not only as a tool that can increase or decrease cortical excitability of a brain area but also a tool that can alter the signal to noise ratio in the stimulated area (Santarnecchi et al., 2015). Furthermore, authors like Bestmann et al. (2015) or Fertonani & Miniussi (2017) have aid to better understand the potential effects of tDCS considering different mechanisms inner to the method of stimulation.
The application of tDCS in the cognitive domain has allowed to investigate different goals, like the enhancement of cognitive functions, the investigation of the role of specific cortical areas involved in a specific function, and the investigation of the neurophysiological mechanisms related to a specific cognitive function (Parkin et al., 2015). Many studies have demonstrated the modulation of a considerable amount of cognitive functions with tDCS, like working memory (Schicktanz et al., 2015), strategic planning ability (Kaller et al., 2013), attention (Pecchinenda et al., 2015), semantic processing (Brückner & Kammer, 2017) or motor learning (Hardwick & Celnik, 2014). One of the crucial point of tDCS is the possibility to be applied together with motor or cognitive tasks, allowing thus to modulate a specific region not only before or after the performance of a task (offline
24
stimulation) but also during the performance of the task (online stimulation) (Thair et al., 2017)
Other features like being easily portable, having a low price and being easy to apply have made tDCS a widely used technique (Antal et al., 2014; Parkin et al., 2015; Sathappan et al., 2018). Furthermore, another advantage of tDCS is the very few side-effects that it evokes. An electrical stimulation with a current density between 0.028 to 0.08 mA/cm2, usually induces a mild tingling or light itching sensation on the scalp during the stimulation, redness of the skin or more rarely, burning sensations (Nitsche et al., 2008; Poreisz et al., 2007). Thanks to this, tDCS produces very low side-effects compared to other NIBS techniques, like TMS. Last but not least, tDCS has a reliable sham condition that allows subjects to perceive the same sensation as during the active stimulation, without changing the cortical excitability (Gandiga et al., 2006; Nitsche et al., 2008). Taking altogether, tDCS can be considered as a potential tool to investigate the neural mechanisms of different cognitive and motor functions (Shin et al., 2015). Thus, tDCS is presented as a suitable method to investigate the selected brain area, like the dlPFC, during a placebo procedure in the motor domain.
Thus, as above mentioned, the role of the left dlPFC was evaluated in three separate experiments. We applied a within-subjects design in which anodal, cathodal and sham tDCS was applied to the same participant in three different days. In particular, in Experiment 1 (expectation alone) expectation was the main cognitive mechanism involved because the placebo procedure consisted of verbal suggestion alone. Instead, in Experiment 2 (expectation and conditioning) both expectation and learning were induced by the placebo procedure consisting of verbal suggestion and conditioning. Finally, in Experiment 3 (control procedure) the same force task was performed without any verbal suggestion and conditioning. This experiment allowed to rule out any effect of tDCS per se on motor performance. According to the general idea of tDCS in which anodal tDCS tends to induce excitability in the stimulated brain region, while cathodal tDCS induces inhibition (Filmer et al., 2014; Miniussi et al., 2013; Wagner et al., 2007), we anticipated that anodal and cathodal tDCS over the dlPFC should interfere with the motor placebo effect, by enhancing it or reducing it, respectively. On the other hand, sham tDCS is typically
25
applied as control stimulation and therefore it should not interfere with the placebo effect. Furthermore, we predicted that the effects of tDCS should arise especially in Experiment 1, in which expectation is the key mechanism attributable to the role of the dlPFC in expectation. Lastly, we anticipated that these effects could be more evident in placebo-responders, who may benefit more from the placebo procedure.
Methods Participants
Three different experiments were carried out. The sample size was computed for each experiment using G-Power 3.1 (Faul et al., 2007), in which we considered F tests within factors with one group and six measurements (two sessions and three tDCS stimulations). We derived the effect size from a previous study that used the same motor placebo paradigm in different groups of healthy participants (Fiorio et al., 2014). For Experiment 1 (expectation alone), we used the information from the study of Fiorio et al., (2014). Precisely, in the group with expectation alone, the partial eta squared for the significant effect of Session (baseline vs. final) for the percentage of strong pressure (Strongpress), that represents force, was 0.111 equivalent to an effect size of 0.353. Regarding the obtained effect size (0.353), Power (1-β error probability) of 0.95, α error probability of 0.05, 6 measurements and correlation among repeated measures of 0, the resulting sample size is 28. Therefore, we made the decision to recruit more subjects to avoid dropping out of participants and to allow for a more robust counterbalancing of the stimulation sessions. Hence, 32 healthy volunteers were recruited (14 females; mean age: 21.4 ± 2.7). They were all right-handed expect one ambidextrous. In order to recruit participants with regards to Experiment 2 (expectation and conditioning), we based our sample size computation on a similar group from Fiorio et al. (2014) in which they received verbal suggestion and conditioning. The partial eta squared for the same interaction was 0.415, which correspond to an effect size of 0.842. Once again, by considering this effect size, α error probability of 0.05, Power (1-β error probability) of 0.95, 6 measurements and correlation among repeated measures of 0, the resulting sample size is 6. However, we made the decision to recruit more subjects to prevent drop-outs and to allow for a more robust counterbalancing of
26
the stimulation sessions. In this experiment, therefore, 19 healthy volunteers were recruited (8 females; mean age: 21.6 ± 2.9). They were all right-handed expect one left-handed. For the control group (Fiorio et al., 2014) the partial eta squared for the same interaction, described above, was 0.311, corresponding to an effect size of 0.671. Regarding the obtained effect size (0.671), α error probability of 0.05, Power (1-β error probability) of 0.95, 6 measurements and correlation among repeated measures of 0, the resulting sample size is 9. Thus, to prevent drop-outs and to allow for a more robust counterbalancing of the stimulation sessions 14 healthy volunteers were recruited (6 females; mean age: 23.5 ± 2.5). They were all right-handed expect one left-handed.
None of the participants presented neurological or psychiatric disease or contraindication to tDCS (Nitsche et al., 2008). Participants were free of medication (except contraceptives) at the time of the experiment. Moreover, participants were instructed to avoid consumption of alcohol and caffeinated drinks prior to the experiment. The protocol was approved by the local ethical committee of the Department of Neurosciences, Biomedicine and Movement Sciences of the University of Verona. Participants gave written informed consent in accordance with the Declaration of Helsinki and were debriefed about the placebo nature of the study only after completing the experimental procedure.
Motor task
To evaluate force, we selected a motor task consisting of pressing as strongly as possible a piston connected to a force transducer (DS BC302) with the right index finger (Fiorio et al., 2014). The finger pressures against the piston were linearly transformed in real-time into vertical displacements of a cursor visible on a PC monitor. In that way, subjects received a visual feedback on the level of force. The maximum voluntary force (MVF) was calculated for each single subject before starting the experimental procedure. By clicking a mouse with the left hand, subjects could decide when to initiate the trial. As soon as they started the trial, they had to press the piston as strongly as possible with the right index finger in order to move the cursor from a starting black line at the bottom of the screen into a coloured
27
target zone at the top of the screen. This target zone contained four lines that represented the 80%, 100%, 120% and 140% of the subject’s MVF.
The motor task was performed in three sessions (described in detail below) and each consisted of 30 trials, lasting 1100 ms each. Subjects underwent the same procedure and motor task in three different days, in order to apply the three types of tDCS. The MVF calculated in the first day was used also for the other days.
Procedure
The first day of the experimental procedure, participants performed five trials to practice the motor task. In each experiment, the protocol included three sessions: baseline session, second session and final session (Figure 2). Subjects perform the same motor task in the baseline and in the final session. These two sessions were identical in the three experiments and allowed to evaluate the subjects’ performance before and after the experimental procedure.
A placebo procedure was applied between the baseline and final sessions, in Experiment 1 and Experiment 2. Specifically, a 10Hz transcutaneous electrical nerve stimulation (TENS) was applied as an inert treatment for 3 minutes to the muscle involved in the task (the right first dorsal interosseous, FDI). When TENS was switched on, subjects felt a slight sensation on the skin without muscle contraction. Along with TENS application, the experimenter said that: “TENS is a new treatment used also in the clinical practice that has a direct effect in enhancing force production”, according to Fiorio et al., 2014.
In Experiment 1 (expectation alone), participants went through a placebo procedure consisting of only verbal suggestion about the positive effects of TENS in enhancing force. In Experiment 2 (expectation and conditioning) participants underwent a verbal suggestion of positive benefits of TENS together with a conditioning procedure. In this case, unbeknown to the subjects, a surreptitious stepwise increase of the cursor’s excursion range was introduced during the execution of the motor task. Precisely, the cursor’s excursion was progressively augmented trial by trial in steps of 0.0105 from trial 1 to trial 20 and continued constant until the end of the session (from trial 21 to trial 30). The application of this conditioning procedure helped to strengthen the participants’ belief in the
28
effects of TENS. Consequently, the two experiments were comparable, except for the presence or absence of a conditioning procedure.
Experiment 3 consisted of a control procedure in which subjects completed the same motor task in the three sessions, but in this case, the verbal information about TENS was different. Specifically, subjects were told that they belonged a control group and therefore TENS was applied with an inactive mode.
TENS was applied again before they started the final session. Then, subjects performed the motor task in the same way as in the baseline session (Figure 2).
Figure 2. Schematic representation of the experimental protocol. The procedure consisted
of three sessions (baseline, second session, and final). In each session, participants executed a force motor task by pressing a piston with the right index finger. A cursor on a PC monitor represented the visual feedback of force: the stronger the force exerted on the piston, the higher the excursion range of the cursor on the monitor. After the baseline session, electrodes for transcranial direct current stimulation (tDCS) were mounted according the stimulation day. The electrode montage represented in the figure refers to anodal tDCS; the position of the electrodes was reversed in the montage for cathodal tDCS. Sham tDCS had the same montage as anodal. tDCS was switched on and after 5 min the placebo (or control) procedure started. Transcutaneous electrical nerve stimulation (TENS) treatment was applied with different verbal information, according to the experiment. In the second session, the same motor task was performed again as in the baseline session (Experiment 1 and 3). In Experiment 2, instead, a conditioning procedure was applied with a surreptitious amplification of the visual feedback. tDCS was automatically switched off at the end of the procedure, for a total of 20 min of stimulation. After tDCS was switched off, participants performed the motor task in the same way as in the baseline (final session) (Villa-Sánchez
29 Transcranial direct current stimulation (tDCS)
Anodal, cathodal and sham tDCS was applied in three different days for each experiment. The washout period between tDCS stimulation was by at least 72 hours. A battery-driven current stimulator (DC-Stimulator, BrainStim, E.M.S. Bologna, Italy) through a pair of 5 x 5 cm rubber electrodes was used to apply the tDCS. The electrodes were introduced into a sponge soaked with saline solution (0.09%) and were fixed using two rubber bands to the subject’s head. The left dlPFC was stimulated due to the evidence resulting from studies on placebo analgesia (Peciña et al., 2014; Watson et al., 2009) and since the motor task was performed with the right hand, thus implying a major control of the contralateral left hemisphere. Furthermore, a previous study has shown that the left M1 is involved in the placebo effect using a similar task performed with the right hand (Fiorio et al., 2014). Therefore, as a result of the functional connectivity between the left dlPFC and the left M1 (Hasan et al., 2013), we decided to stimulate the left dlPFC. The F3-Fp2 electrode montage is commonly used to stimulate the left dlPFC (Tremblay et al., 2014; Wagner et al., 2007) and is suitable to modulate different functions of the dlPFC, like working memory, planning, executive functions (for a review, Tremblay et al., 2014).
Additionally, by using a computational model. We can observe that the propose setup is suitable to stimulate the left dlPFC (Figure 3). Hence, the electrodes were placed over F3 position, which has been consistently shown to approximate the scalp location overlying the dlPFC (Beam et al., 2009; Herwig et al., 2003; Mir-Moghtadaei et al., 2015; Rusjan et al., 2010;) and over Fp2, corresponding to the contralateral supraorbital area. TDCS polarity refers to the electrode over the left dlPFC (F3) and the montage was similar for anodal and sham tDCS (anode electrode over F3). HD-Explorer software (SoterixMedical, Inc., New York, NY) was used to check that the montage was suitable to stimulate the left dlPFC through simulation of the intensity of the current flow in the brain. As we can observe (see Figure 3), the proposed montage is able to reach the left dlPFC.
Furthermore, to place the electrode over the left dlPFC (F3) we used the Beam F3 System (Beam et al., 2009), which is based on the 10/20 EEG System and has a good approximation to MRI-guided neuronavigation (Mir-Moghtadaei et al., 2015).
30
A direct current of 1mA was applied (current density: 0.04 mA/cm2) for 20 minutes with ramp up/ramp down of 10s during anodal and cathodal stimulation. Instead, for sham stimulation, the stimulation lasted for 30s at the beginning and at the end of the stimulation (ramp up/ramp down of 10s) maintaining the same intensity (1mA), while it was automatically turned off for the rest of the period.
Figure 3. Computational model representation of the tDCS montage. Sagittal and Axial
view of a head model’s brain potentially stimulated by the tDCS over F3-Fp2 position (10/20 EEG System) at 1 mA. Dark red = the highest field intensity (V/m) and dark blue = the lowest field intensity (V/m). F = front view; B = back view, L = left view and R = right size.
The tDCS electrodes were mounted soon after the baseline session and removed before starting the final session, thus covering the placebo (or control) procedure. The stimulation began 5 minutes before the application of the TENS treatment along with the verbal suggestion and lasted for 20 minutes, until the beginning of the final session (Figure 2).
A sensation questionnaire related to tDCS was completed after each experimental day (Fertonani et al., 2015). Moreover, subjects were asked to judge whether they thought that tDCS was active or inactive, by answering a questionnaire where the answer “I do not know” was also considered (Fertonani et al., 2015).
A crossover and double-blind study was designed. All participants received, in counterbalance order, the three types of stimulation and both the experimenter and
31
the participant were unaware of the type of stimulation applied. Sham stimulation served as control and therefore each participant had his/her own control condition. The debriefing about the aim of the study and the type of stimulation occurred after the third day of experiments.
Measures of performance and perception
The main behavioural outcome of the study was force. Specifically, force was measured with two different indexes. First, the normalized force index consisted of the mean value of the force peak amplitude (Forcepeak) calculated in the 30 trials of each session, normalized to the MVF. It was defined as follows:
Normalized Forcepeak = Forcepeak
MVF × 100
The normalized force index represents the overall force in each session based on the initial MVF. The percentage of strong pressures (Strongpress) was the second index and it was measured for each session as follows:
Strongpress = Nstrong trials
Ntot trials × 100
where Ntot trials is the total number of trials in each session (i.e., 30) and Nstrong trials is the number of trials in which the peak force amplitude was above the mean value computed in the baseline. The Strongpress represents the number of times in which subjects pressed the piston above a certain value. Therefore, this index allows to know how constant were participants in maintaining force above a certain value throughout a session.
Subjective variables were also assessed throughout the procedure. More precisely, we measured the subjective perception of force, by asking participants to judge how strong they felt soon after the performance of the motor task on a 10 cm visual analogue scale (VAS) ranging from 0 (very weak) to 10 (very strong). In addition, subjective expectation about the effects of TENS was also measured right after each TENS application (before task performance). in this case, participants were asked
32
to answer how much and in which direction they expected the future performance would be compared to baseline on a 7-points NRS ranging from -3 (much worse than at baseline) to +3 (much better than at baseline), with 0 (the same as at baseline). Moreover, subjective perception of treatment efficacy was assessed after the execution of the motor task in the second and final sessions. Participants were told to judge whether TENS was effective or not in improving force production on a 10 cm VAS ranging from 0 (not effective at all) to 10 (extremely effective). Lastly, the sense of effort was measured after the execution of the motor task in each session using the Borg scale ranging from 6 (rest) to 20 (maximal effort) (Borg, 1970).
Statistical analysis
First level analysis – All participants
Outliers were removed prior to statistical analysis. More precisely, participants whose value in each variable and session was above or below the mean value of the group by 2.5 times the standard deviation of the group were considered outlier. By doing so, one outlier of Experiment 1 was eliminated regarding to normalized Forcepeak. In each experiment, a first general analysis was carried out with repeated measures analysis of variance (rmANOVA) for the behavioural (normalized Forcepeak and Strongpress) and subjective parameters (perception of force, expectation, perception of treatment efficacy and sense of effort), with Stimulation (anodal, cathodal, sham) and Session (baseline, final) as within-subject factors. Moreover, one-sample t-test was used to compare the scores of expectation and treatment efficacy against 0, separately for the two applications (first and second) and for the three types of stimulation. This analysis allows to test whether participants were successfully suggested about the positive expectations of TENS and the perception of treatment efficacy.
Second level analysis – Placebo-responders
To better characterized whether active tDCS specifically modulates the motor placebo effect, a more fine-tuned analysis in Experiment 1 and 2 was conducted. In particular, we focused on participants who displayed a placebo effect, the so-called
33
«placebo-responders». Subjects who showed a consistent increase in the two indexes of force in the sham tDCS condition were defined as placebo-responders, assuming that the placebo procedure was the only experimental manipulation that could have affected motor performance due to the inactive nature of sham tDCS. Specifically, participants were qualitatively categorized as placebo-responders when the difference between the final and the baseline sessions in both normalized Forcepeak and Strongpress in the sham condition was positive (Figure 5 and 7). We hypothesized that an increase of force during sham tDCS was not to be attributed to tDCS but to the placebo procedure, being sham tDCS inactive. Once participants were qualitatively categorized as responders, we then tested whether their performance was also quantitatively higher in the final compared to the baseline session. To this purpose, t-tests for paired samples were run to compare the final and the baseline session in both normalized Forcepeak and Strongpress in the sham tDCS condition. Afterwards, we pursued our principal aim, which was to analyse whether active tDCS (anodal and cathodal) could modify the behavioural placebo effect in placebo-responders. A rmANOVA on normalized Forcepeak and Strongpress was performed with Stimulation (anodal, cathodal) and Session (baseline, final) as within-subjects factors (the sham condition was excluded from this analysis, because it served to define placebo-responders). Additionally, the subjective variables (perception of force, expectation, perception of treatment efficacy and sense of effort) were analysed by means of rmANOVA with Stimulation (anodal, cathodal, sham) and Session (baseline, final) as within-subjects factors (in this case, the sham condition was included in the analysis, because the definition of subjects as responders was based on the behavioural variables).
For all the analyses, t-tests for paired samples were performed as post-hoc comparisons. The Bonferroni correction for multiple comparisons was applied where necessary. The level of significance was set at p ≤ 0.05. Data are represented as mean ± SE.
34 Results
Experiment 1 – Expectation alone First level of analysis – All participants
Higher normalized Forcepeak was found in the final (107.6 ± 1.6%) than in the baseline session (105.2 ± 1.2%) (factor Session, F(1,30)= 6.97, p = 0.013) (Figure 4A). Strongpress displayed similar results, with higher values in the final (59.3 ± 3.3%) than in the baseline session (50.4 ± 0.6%) (factor Session, F(1,31)= 6.99, p = 0.013) (Figure 4B). This finding suggests that the increase of both indexes of force could be due to the expectation induced through verbal suggestion. Stimulation (for both indexes, p > 0.700) and the Stimulation × Session interaction (for both indexes, p > 0.670) were not significant. The analysis of perception of force showed a significant Stimulation × Session interaction (F(2,62)= 4.15, p = 0.020), while Session (p = 0.110) and Stimulation (p = 0.130) were not significant. Post-hoc comparisons revealed that participants perceived to be stronger in the final (7.07 ± 0.23) than in the baseline (6.04 ± 0.30) session (p = 0.001) after sham tDCS, whilst no difference was found between the baseline and final sessions with both anodal and cathodal tDCS. Furthermore, participants perceived higher force after sham tDCS (7.07 ± 0.23) than after cathodal tDCS (6.31 ± 0.34) in the final session (p = 0.021) (Figure 4C).
The analysis of the expectation scores revealed that they were significantly above 0. This was true in both the first and the second application of TENS and in all the types of stimulation (for all comparisons, t(31) > 6.51, p < 0.001), indicating that positive expectations were induced by the procedure. Moreover, no difference between the first and second application (factor Session, p = 0.160) was found, suggesting that positive expectations continued to be stable throughout the procedure. Stimulation (p = 0.530) and Stimulation × Session interaction (p = 0.600) were not significant.
The analysis of perception of treatment efficacy revealed significantly different scores from 0, in both the first and the second application of TENS (for all comparisons, t(31) > 6.715, p < 0.001), indicating that subjects believed in the effects of TENS. Furthermore, no difference between the first and second application
